Create a chat technical support for the site

We are seeking an experienced web developer to re-engineer and fix critical bugs in an existing web application. The application was developed using pure JavaScript (Vanilla JS), without frameworks, and is deployed on Vercel. Currently, the application is functional but experiences intermittent crashes and freezes in certain functionalities that require urgent attention. The application's technical stack includes: - Frontend: HTML5, CSS3, ES6+ JavaScript. - Backend/Data: Supabase (using PostgreSQL, Auth, and Realtime) and Upstash (for Redis). - Deployment: Vercel. The main objective of this project is to identify and resolve the root causes of the problems, as well as optimize the code to improve stability and performance. Key tasks include: 1. Thorough analysis of network errors and logs available in Supabase and Upstash to pinpoint the source of the failures. 2. Refactoring the JavaScript code to ensure robust handling of loading and error states. 3. Optimizing state persistence on the frontend. 4. Improving communication between the client and Supabase/Upstash services, including query efficiency and secure API integration. The current code, although generated with AI assistance (Claude Code), has a logical and modular structure, but requires human review and expertise to achieve the necessary quality and robustness standards. The developer is expected to have a solid understanding of modern JavaScript, relational and non-relational databases, and cloud deployment platforms.

Looking for a developer to create an AI assistant based on OpenAI GPT-5.5 with integration of Tourvisor API (Tour Search).What needs to be implementedBackend Preferably: Python (FastAPI) or Node.js The backend must have a REST API for integration with external services.Integration with OpenAI It is necessary to use OpenAI API (GPT-5.5). The AI must: conduct a dialogue with the client; remember the conversation history; automatically determine the language (Russian/Kazakh); respond as a travel agency manager; use a system prompt and knowledge base.Integration with Tourvisor At the first stage, only: API "Tour Search" After receiving data from the client, the backend should call the Tourvisor API and get real offers. The architecture should allow for subsequent connection: API "Hot Tours"; API "Hotel Descriptions"; without significant reworking of the project.Logic of operation The AI must independently conduct the dialogue. For example: Client: I want to go to Turkey The AI clarifies: departure city; dates; number of adults; number of children; ages of children; budget; wishes. After receiving the necessary data, the backend makes a request to the Tourvisor API. The results obtained are passed to GPT to form a nice response to the client.Important requirements The AI must not: make up prices; make up availability; make up tours. Use only information obtained from the Tourvisor API.Function Calling (Tools) is desirable It is necessary to implement separate functions: searchTours() In the future, the following will be added: searchHotTours() getHotelInfo()Admin panel A simple web panel is needed where you can: change the system prompt; upload/change the knowledge base; view conversation history; view logs and errors.Technologies Python FastAPI (preferably) Docker PostgreSQL Swagger/OpenAPIWhat should be delivered upon completion source code of the project; installation instructions; Docker configuration; API documentation; assistance with the first launch.Will be an advantage experience with OpenAI API; experience integrating with Tourvisor API; experience creating AI assistants for WhatsApp, Telegram, or other messengers.Additional wish The project should be written modularly so that in the future it can be connected without reworking: WhatsApp (Talk Me or another service); Telegram; Instagram Direct; company website. Important: provide for the possibility of connecting the API "Hot Tours" and "Hotel Descriptions" in the future without changing the main logic of the application.

Using my API data and wallet, create a Python script that successfully places a test order on the exchange.

A simple Python script is needed for Windows that works with Microsoft Edge, opens my LinkedIn feed, scrolls through it, finds posts with specific words, and writes the results to a file. We had a working script, but a few days ago it stopped working due to an Edge update; it just no longer automates. What the script should do 0. Read the list of keywords and the number of posts to process from the configuration. 1. Open Microsoft Edge. 2. Go to: https://www.linkedin.com/feed/ 3. I log in manually if needed. 4. The script scrolls down the feed, for example, 30 posts. 5. It searches for the specified words in the posts. 6. If it finds a post with the required word, it writes to CSV or JSON: post_url, author, text, matched_keyword Doing it in Python and running it in Edge are mandatory conditions.

Need an experienced Python developer to create an automation program for working with the website. What the bot should do: Log in to my account on the website. Continuously (24/7) monitor for new Work Orders. Upon the appearance of a new order, open its card. Analyze the information about the order (type of work, city, distance, description). Automatically accept orders that meet the specified conditions. If the order does not meet the conditions, skip it. Send notifications in Telegram about accepted and skipped orders. Requirements: Python. Playwright (preferred) or Selenium. Work through a regular browser. Transfer the complete source code after the project is completed. Provide installation and launch instructions. Mandatory experience in developing similar automation systems. When responding, please send examples of similar projects.